Markerless tracking for mobile augmented reality

This work aims to realize a recognition system for markerless tracking in mobile augmented reality. A method based local invariant descriptors is implemented to extract image feature points for natural fiducial identification. This technique is optimized and adapted for a mobile architecture with low resources and deployed on a portable device. Afterwards, a hybrid approach of camera pose estimation is proposed to augment real images with virtual graphics. Experiments and many evaluations are conducted to demonstrate the relevance of the proposed approach. This study enabled to overcome some identification issues for markerless tracking in mobile augmented reality environments.